Artificial Intelligence Nanodegree¶
Convolutional Neural Networks¶
Project: Write an Algorithm for a Dog Identification App¶
In this notebook, some template code has already been provided for you, and you will need to implement additional functionality to successfully complete this project. You will not need to modify the included code beyond what is requested. Sections that begin with '(IMPLEMENTATION)' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!
Note: Once you have completed all of the code implementations, you need to finalize your work by exporting the iPython Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You can then export the notebook by using the menu above and navigating to \n", "File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.
In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question X' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.
Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. Markdown cells can be edited by double-clicking the cell to enter edit mode.
The rubric contains optional "Stand Out Suggestions" for enhancing the project beyond the minimum requirements. If you decide to pursue the "Stand Out Suggestions", you should include the code in this IPython notebook.
Why We're Here¶
In this notebook, you will make the first steps towards developing an algorithm that could be used as part of a mobile or web app. At the end of this project, your code will accept any user-supplied image as input. If a dog is detected in the image, it will provide an estimate of the dog's breed. If a human is detected, it will provide an estimate of the dog breed that is most resembling. The image below displays potential sample output of your finished project (... but we expect that each student's algorithm will behave differently!).
In this real-world setting, you will need to piece together a series of models to perform different tasks; for instance, the algorithm that detects humans in an image will be different from the CNN that infers dog breed. There are many points of possible failure, and no perfect algorithm exists. Your imperfect solution will nonetheless create a fun user experience!
The Road Ahead¶
We break the notebook into separate steps. Feel free to use the links below to navigate the notebook.
- Step 0: Import Datasets
- Step 1: Detect Humans
- Step 2: Detect Dogs
- Step 3: Create a CNN to Classify Dog Breeds (from Scratch)
- Step 4: Use a CNN to Classify Dog Breeds (using Transfer Learning)
- Step 5: Create a CNN to Classify Dog Breeds (using Transfer Learning)
- Step 6: Write your Algorithm
- Step 7: Test Your Algorithm
Step 0: Import Datasets¶
Import Dog Dataset¶
In the code cell below, we import a dataset of dog images. We populate a few variables through the use of the load_files function from the scikit-learn library:
train_files,valid_files,test_files- numpy arrays containing file paths to imagestrain_targets,valid_targets,test_targets- numpy arrays containing onehot-encoded classification labelsdog_names- list of string-valued dog breed names for translating labels
from sklearn.datasets import load_files
from keras.utils import np_utils
import numpy as np
from glob import glob
# define function to load train, test, and validation datasets
def load_dataset(path):
data = load_files(path)
dog_files = np.array(data['filenames'])
dog_targets = np_utils.to_categorical(np.array(data['target']), 133)
return dog_files, dog_targets
# load train, test, and validation datasets
train_files, train_targets = load_dataset('dogImages/train')
valid_files, valid_targets = load_dataset('dogImages/valid')
test_files, test_targets = load_dataset('dogImages/test')
# load list of dog names
dog_names = [item[20:-1] for item in sorted(glob("dogImages/train/*/"))]
# print statistics about the dataset
print('There are %d total dog categories.' % len(dog_names))
print('There are %s total dog images.\n' % len(np.hstack([train_files, valid_files, test_files])))
print('There are %d training dog images.' % len(train_files))
print('There are %d validation dog images.' % len(valid_files))
print('There are %d test dog images.'% len(test_files))
Import Human Dataset¶
In the code cell below, we import a dataset of human images, where the file paths are stored in the numpy array human_files.
import random
random.seed(8675309)
# load filenames in shuffled human dataset
human_files = np.array(glob("lfw/*/*"))
random.shuffle(human_files)
# print statistics about the dataset
print('There are %d total human images.' % len(human_files))
Step 1: Detect Humans¶
We use OpenCV's implementation of Haar feature-based cascade classifiers to detect human faces in images. OpenCV provides many pre-trained face detectors, stored as XML files on github. We have downloaded one of these detectors and stored it in the haarcascades directory.
In the next code cell, we demonstrate how to use this detector to find human faces in a sample image.
import cv2
import matplotlib.pyplot as plt
%matplotlib inline
# extract pre-trained face detector
face_cascade = cv2.CascadeClassifier('haarcascades/haarcascade_frontalface_alt.xml')
# load color (BGR) image
img = cv2.imread(human_files[3])
# convert BGR image to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# find faces in image
faces = face_cascade.detectMultiScale(gray)
# print number of faces detected in the image
print('Number of faces detected:', len(faces))
# get bounding box for each detected face
for (x,y,w,h) in faces:
# add bounding box to color image
cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)
# convert BGR image to RGB for plotting
cv_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# display the image, along with bounding box
plt.imshow(cv_rgb)
plt.show()
Before using any of the face detectors, it is standard procedure to convert the images to grayscale. The detectMultiScale function executes the classifier stored in face_cascade and takes the grayscale image as a parameter.
In the above code, faces is a numpy array of detected faces, where each row corresponds to a detected face. Each detected face is a 1D array with four entries that specifies the bounding box of the detected face. The first two entries in the array (extracted in the above code as x and y) specify the horizontal and vertical positions of the top left corner of the bounding box. The last two entries in the array (extracted here as w and h) specify the width and height of the box.
Write a Human Face Detector¶
We can use this procedure to write a function that returns True if a human face is detected in an image and False otherwise. This function, aptly named face_detector, takes a string-valued file path to an image as input and appears in the code block below.
# returns "True" if face is detected in image stored at img_path
def face_detector(img_path):
img = cv2.imread(img_path)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
faces = face_cascade.detectMultiScale(gray)
return len(faces) > 0
(IMPLEMENTATION) Assess the Human Face Detector¶
Question 1: Use the code cell below to test the performance of the face_detector function.
- What percentage of the first 100 images in
human_fileshave a detected human face? - What percentage of the first 100 images in
dog_fileshave a detected human face?
Ideally, we would like 100% of human images with a detected face and 0% of dog images with a detected face. You will see that our algorithm falls short of this goal, but still gives acceptable performance. We extract the file paths for the first 100 images from each of the datasets and store them in the numpy arrays human_files_short and dog_files_short.
Answer:
- Percentage of human_files_short where a human face has been detected: 99.0
- Percentage of dog_files_short where a human face has been detected: 11.0
human_files_short = human_files[:100]
dog_files_short = train_files[:100]
# Do NOT modify the code above this line.
## TODO: Test the performance of the face_detector algorithm
## on the images in human_files_short and dog_files_short.
def get_percentage_detected_faces(img_path_list):
return 100 * sum([face_detector(img_path) for img_path in img_path_list]) \
/ len(img_path_list)
def print_message_percentage_detected_faces(img_group_name_string):
print("Percentage of {} where a human face has been detected: {}".\
format(img_group_name_string,
get_percentage_detected_faces(eval(img_group_name_string))))
print_message_percentage_detected_faces('human_files_short')
print_message_percentage_detected_faces('dog_files_short')
Question 2: This algorithmic choice necessitates that we communicate to the user that we accept human images only when they provide a clear view of a face (otherwise, we risk having unneccessarily frustrated users!). In your opinion, is this a reasonable expectation to pose on the user? If not, can you think of a way to detect humans in images that does not necessitate an image with a clearly presented face?
Answer: In an ideal world, one could argue that users need not know the inner workings of the tools they use. It is the developer's responsibility to make sure a software application, for example, is able to handle satisfactorily any input. But such an approach is guaranteed to fail as long as tools are imperfect. Consequently, users may need to understand that the app they are planning to use is not magic, and as such the technology it is based on has its limitations. As with everything we use - be it a mechanical device, a piece of software, or a service -, we are bound to get better results if we understand its workings, at least a high level. This understanding also helps set realistic expectations and therefore avoid frustration with disappointing results. I therefore personally tend to favour this second perspective: yes, it is reasonable to make users aware of those constraints on the input they provide which may impact on the quality of the app's output. --- As for the method itself, Haar cascades seem to be a good technique for detecting objects. If what we are after is human faces, as is the case in this project (where we want to find a resemblance with a dog breed based on a human's face), then OpenCV's pretrained Haar cascade face detectors seem a reasonably well performing method: they give almost no false negatives on our small sample set, but do get some false positives because some of the features they use (eyes, etc.) are also found in animal faces. However, if we want to detect humans, not just faces, we would need to train our own Haar cascade human detectors on an appropriate dataset, with positive (humans, not necessarily human faces) and negative (no humans) examples. OpenCV do provide a pretrained full body detector too.
We suggest the face detector from OpenCV as a potential way to detect human images in your algorithm, but you are free to explore other approaches, especially approaches that make use of deep learning :). Please use the code cell below to design and test your own face detection algorithm. If you decide to pursue this optional task, report performance on each of the datasets.
## (Optional) TODO: Report the performance of another
## face detection algorithm on the LFW dataset
### Feel free to use as many code cells as needed.
Step 2: Detect Dogs¶
In this section, we use a pre-trained ResNet-50 model to detect dogs in images. Our first line of code downloads the ResNet-50 model, along with weights that have been trained on ImageNet, a very large, very popular dataset used for image classification and other vision tasks. ImageNet contains over 10 million URLs, each linking to an image containing an object from one of 1000 categories. Given an image, this pre-trained ResNet-50 model returns a prediction (derived from the available categories in ImageNet) for the object that is contained in the image.
from keras.applications.resnet50 import ResNet50
# define ResNet50 model
ResNet50_model = ResNet50(weights='imagenet')
Pre-process the Data¶
When using TensorFlow as backend, Keras CNNs require a 4D array (which we'll also refer to as a 4D tensor) as input, with shape
$$ (\text{nb_samples}, \text{rows}, \text{columns}, \text{channels}), $$where nb_samples corresponds to the total number of images (or samples), and rows, columns, and channels correspond to the number of rows, columns, and channels for each image, respectively.
The path_to_tensor function below takes a string-valued file path to a color image as input and returns a 4D tensor suitable for supplying to a Keras CNN. The function first loads the image and resizes it to a square image that is $224 \times 224$ pixels. Next, the image is converted to an array, which is then resized to a 4D tensor. In this case, since we are working with color images, each image has three channels. Likewise, since we are processing a single image (or sample), the returned tensor will always have shape
The paths_to_tensor function takes a numpy array of string-valued image paths as input and returns a 4D tensor with shape
Here, nb_samples is the number of samples, or number of images, in the supplied array of image paths. It is best to think of nb_samples as the number of 3D tensors (where each 3D tensor corresponds to a different image) in your dataset!
from keras.preprocessing import image
from tqdm import tqdm
def path_to_tensor(img_path):
# loads RGB image as PIL.Image.Image type
img = image.load_img(img_path, target_size=(224, 224))
# convert PIL.Image.Image type to 3D tensor with shape (224, 224, 3)
x = image.img_to_array(img)
# convert 3D tensor to 4D tensor with shape (1, 224, 224, 3) and return 4D tensor
return np.expand_dims(x, axis=0)
def paths_to_tensor(img_paths):
list_of_tensors = [path_to_tensor(img_path) for img_path in tqdm(img_paths)]
return np.vstack(list_of_tensors)
Making Predictions with ResNet-50¶
Getting the 4D tensor ready for ResNet-50, and for any other pre-trained model in Keras, requires some additional processing. First, the RGB image is converted to BGR by reordering the channels. All pre-trained models have the additional normalization step that the mean pixel (expressed in RGB as $[103.939, 116.779, 123.68]$ and calculated from all pixels in all images in ImageNet) must be subtracted from every pixel in each image. This is implemented in the imported function preprocess_input. If you're curious, you can check the code for preprocess_input here.
Now that we have a way to format our image for supplying to ResNet-50, we are now ready to use the model to extract the predictions. This is accomplished with the predict method, which returns an array whose $i$-th entry is the model's predicted probability that the image belongs to the $i$-th ImageNet category. This is implemented in the ResNet50_predict_labels function below.
By taking the argmax of the predicted probability vector, we obtain an integer corresponding to the model's predicted object class, which we can identify with an object category through the use of this dictionary.
from keras.applications.resnet50 import preprocess_input, decode_predictions
def ResNet50_predict_labels(img_path):
# returns prediction vector for image located at img_path
img = preprocess_input(path_to_tensor(img_path))
return np.argmax(ResNet50_model.predict(img))
Write a Dog Detector¶
While looking at the dictionary, you will notice that the categories corresponding to dogs appear in an uninterrupted sequence and correspond to dictionary keys 151-268, inclusive, to include all categories from 'Chihuahua' to 'Mexican hairless'. Thus, in order to check to see if an image is predicted to contain a dog by the pre-trained ResNet-50 model, we need only check if the ResNet50_predict_labels function above returns a value between 151 and 268 (inclusive).
We use these ideas to complete the dog_detector function below, which returns True if a dog is detected in an image (and False if not).
### returns "True" if a dog is detected in the image stored at img_path
def dog_detector(img_path):
prediction = ResNet50_predict_labels(img_path)
return ((prediction <= 268) & (prediction >= 151))
(IMPLEMENTATION) Assess the Dog Detector¶
Question 3: Use the code cell below to test the performance of your dog_detector function.
- What percentage of the images in
human_files_shorthave a detected dog? - What percentage of the images in
dog_files_shorthave a detected dog?
Answer:
- Percentage of human_files_short where a dog has been detected: 1.0
- Percentage of dog_files_short where a dog has been detected: 100.0
### TODO: Test the performance of the dog_detector function
### on the images in human_files_short and dog_files_short.
def get_percentage_detected_dogs(img_path_list):
return 100 * sum([dog_detector(img_path) for img_path in img_path_list]) \
/ len(img_path_list)
def print_message_percentage_detected_dogs(img_group_name_string):
print("Percentage of {} where a dog has been detected: {}".\
format(img_group_name_string,
get_percentage_detected_dogs(eval(img_group_name_string))))
print_message_percentage_detected_dogs('human_files_short')
print_message_percentage_detected_dogs('dog_files_short')
Step 3: Create a CNN to Classify Dog Breeds (from Scratch)¶
Now that we have functions for detecting humans and dogs in images, we need a way to predict breed from images. In this step, you will create a CNN that classifies dog breeds. You must create your CNN from scratch (so, you can't use transfer learning yet!), and you must attain a test accuracy of at least 1%. In Step 5 of this notebook, you will have the opportunity to use transfer learning to create a CNN that attains greatly improved accuracy.
Be careful with adding too many trainable layers! More parameters means longer training, which means you are more likely to need a GPU to accelerate the training process. Thankfully, Keras provides a handy estimate of the time that each epoch is likely to take; you can extrapolate this estimate to figure out how long it will take for your algorithm to train.
We mention that the task of assigning breed to dogs from images is considered exceptionally challenging. To see why, consider that even a human would have great difficulty in distinguishing between a Brittany and a Welsh Springer Spaniel.
| Brittany | Welsh Springer Spaniel |
|---|---|
 |
 |
It is not difficult to find other dog breed pairs with minimal inter-class variation (for instance, Curly-Coated Retrievers and American Water Spaniels).
| Curly-Coated Retriever | American Water Spaniel |
|---|---|
 |
 |
Likewise, recall that labradors come in yellow, chocolate, and black. Your vision-based algorithm will have to conquer this high intra-class variation to determine how to classify all of these different shades as the same breed.
| Yellow Labrador | Chocolate Labrador | Black Labrador |
|---|---|---|
 |
 |
 |
We also mention that random chance presents an exceptionally low bar: setting aside the fact that the classes are slightly imabalanced, a random guess will provide a correct answer roughly 1 in 133 times, which corresponds to an accuracy of less than 1%.
Remember that the practice is far ahead of the theory in deep learning. Experiment with many different architectures, and trust your intuition. And, of course, have fun!
Pre-process the Data¶
We rescale the images by dividing every pixel in every image by 255.
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True
# pre-process the data for Keras
train_tensors = paths_to_tensor(train_files).astype('float32')/255
valid_tensors = paths_to_tensor(valid_files).astype('float32')/255
test_tensors = paths_to_tensor(test_files).astype('float32')/255
print(train_tensors.shape)
print(valid_tensors.shape)
print(test_tensors.shape)
(IMPLEMENTATION) Model Architecture¶
Create a CNN to classify dog breed. At the end of your code cell block, summarize the layers of your model by executing the line:
model.summary()
We have imported some Python modules to get you started, but feel free to import as many modules as you need. If you end up getting stuck, here's a hint that specifies a model that trains relatively fast on CPU and attains >1% test accuracy in 5 epochs:
Question 4: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. If you chose to use the hinted architecture above, describe why you think that CNN architecture should work well for the image classification task.
Answer: The proposed architecture seems to me powerful enough to work reasonably well on the task at hand. Concretely, the sequence of three convolution-max pooling pairs of layers followed by a global average pooling (GAP) layer gradually increases the depth and reduces the height and width of the features. This gives the network the ability to discover in the input images increasingly complex local patterns (from low-level ones, such as edges, to higher and higher level ones, such as shapes, then specific objects). I would have liked to experiment with adding a few more conv-pooling layer pairs, but I found myself dramatically limited by the computational power of my machine (I was unable to use GPU on AWS due to the very poor Internet connection I have access to these days); as it is, training this network on my CPU took hours; this limitation is extremely frustrating, as it means experimenting with deep learning architectures (unless we do transfer learning and use pretrained parameters) is conditioned on the hardware one can access. --- Unlike in the proposed architecture, I applied dropout to two layers: the last max-pooling layer and the GAP layer. While helping reduce the risk of overfit, this should also speed up the training.
from keras.layers import Conv2D, MaxPooling2D, GlobalAveragePooling2D
from keras.layers import Dropout, Flatten, Dense
from keras.models import Sequential
model = Sequential()
### TODO: Define your architecture.
model.add(Conv2D(filters=16, kernel_size=2, padding='same', activation='relu',
input_shape=(224, 224, 3)))
model.add(MaxPooling2D(pool_size=2))
model.add(Conv2D(filters=32, kernel_size=2, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Conv2D(filters=64, kernel_size=2, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Dropout(0.3))
model.add(GlobalAveragePooling2D())
model.add(Dropout(0.4))
model.add(Dense(133, activation='softmax'))
model.summary()
Compile the Model¶
model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])
(IMPLEMENTATION) Train the Model¶
Train your model in the code cell below. Use model checkpointing to save the model that attains the best validation loss.
You are welcome to augment the training data, but this is not a requirement.
STUDENT'S NOTE: In addition to the checkpointer, I have set an early stopping monitor to limit the training time.
from keras.callbacks import ModelCheckpoint, EarlyStopping
### TODO: specify the number of epochs that you would like to use to train the model.
# Thanks to an early stopping monitor, we can allow setting a high number of epochs. If no improvement in the validation loss is recorded over a prespecified number of epoches (the 'patience' argument to EarlyStopping), then the training will stop. So the training will go on until the first of the two happens: either 'epochs' epochs, or at least 'patience' epochs with no improvement in the validation loss.
epochs = 100
# Set early stopping monitor.
early_stopping_monitor = EarlyStopping(patience=3)
### Do NOT modify the code below this line. --- I have nevertheless set an early stopping monitor to limit the training time.
checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.from_scratch.hdf5',
verbose=1, save_best_only=True)
model.fit(train_tensors, train_targets,
validation_data=(valid_tensors, valid_targets),
epochs=epochs, batch_size=20,
callbacks=[checkpointer, early_stopping_monitor],
verbose=1)
Load the Model with the Best Validation Loss¶
model.load_weights('saved_models/weights.best.from_scratch.hdf5')
Test the Model¶
Try out your model on the test dataset of dog images. Ensure that your test accuracy is greater than 1%.
# get index of predicted dog breed for each image in test set
dog_breed_predictions = [np.argmax(model.predict(np.expand_dims(tensor, axis=0))) \
for tensor in test_tensors]
# report test accuracy
test_accuracy = 100*np.sum(np.array(dog_breed_predictions)==
np.argmax(test_targets, axis=1))/len(dog_breed_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
bottleneck_features = np.load('bottleneck_features/DogVGG16Data.npz')
train_VGG16 = bottleneck_features['train']
valid_VGG16 = bottleneck_features['valid']
test_VGG16 = bottleneck_features['test']
Model Architecture¶
The model uses the the pre-trained VGG-16 model as a fixed feature extractor, where the last convolutional output of VGG-16 is fed as input to our model. We only add a global average pooling layer and a fully connected layer, where the latter contains one node for each dog category and is equipped with a softmax.
VGG16_model = Sequential()
VGG16_model.add(GlobalAveragePooling2D(input_shape=train_VGG16.shape[1:]))
VGG16_model.add(Dense(133, activation='softmax'))
VGG16_model.summary()
Compile the Model¶
VGG16_model.compile(loss='categorical_crossentropy', optimizer='rmsprop',
metrics=['accuracy'])
Train the Model¶
checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.VGG16.hdf5',
verbose=1, save_best_only=True)
VGG16_model.fit(train_VGG16, train_targets,
validation_data=(valid_VGG16, valid_targets),
epochs=20, batch_size=20,
callbacks=[checkpointer, early_stopping_monitor], verbose=1)
Load the Model with the Best Validation Loss¶
VGG16_model.load_weights('saved_models/weights.best.VGG16.hdf5')
Test the Model¶
Now, we can use the CNN to test how well it identifies breed within our test dataset of dog images. We print the test accuracy below.
# get index of predicted dog breed for each image in test set
VGG16_predictions = [np.argmax(VGG16_model.predict(np.expand_dims(feature, axis=0))) \
for feature in test_VGG16]
# report test accuracy
test_accuracy = 100*np.sum(np.array(VGG16_predictions)==
np.argmax(test_targets, axis=1))/len(VGG16_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
Predict Dog Breed with the Model¶
from extract_bottleneck_features import *
def VGG16_predict_breed(img_path):
# extract bottleneck features
bottleneck_feature = extract_VGG16(path_to_tensor(img_path))
# obtain predicted vector
predicted_vector = VGG16_model.predict(bottleneck_feature)
# return dog breed that is predicted by the model
return dog_names[np.argmax(predicted_vector)]
# test
VGG16_predict_breed("dogImages/test/004.Akita/Akita_00276.jpg")
# test
VGG16_predict_breed("dogImages/test/054.Collie/Collie_03835.jpg")
# test
VGG16_predict_breed("dogImages/test/072.German_shorthaired_pointer/German_shorthaired_pointer_04971.jpg")
Step 5: Create a CNN to Classify Dog Breeds (using Transfer Learning)¶
You will now use transfer learning to create a CNN that can identify dog breed from images. Your CNN must attain at least 60% accuracy on the test set.
In Step 4, we used transfer learning to create a CNN using VGG-16 bottleneck features. In this section, you must use the bottleneck features from a different pre-trained model. To make things easier for you, we have pre-computed the features for all of the networks that are currently available in Keras:
- VGG-19 bottleneck features
- ResNet-50 bottleneck features
- Inception bottleneck features
- Xception bottleneck features
The files are encoded as such:
Dog{network}Data.npz
where {network}, in the above filename, can be one of VGG19, Resnet50, InceptionV3, or Xception. Pick one of the above architectures, download the corresponding bottleneck features, and store the downloaded file in the bottleneck_features/ folder in the repository.
(IMPLEMENTATION) Obtain Bottleneck Features¶
In the code block below, extract the bottleneck features corresponding to the train, test, and validation sets by running the following:
bottleneck_features = np.load('bottleneck_features/Dog{network}Data.npz')
train_{network} = bottleneck_features['train']
valid_{network} = bottleneck_features['valid']
test_{network} = bottleneck_features['test']### TODO: Obtain bottleneck features from another pre-trained CNN.
bottleneck_features = np.load('bottleneck_features/DogResnet50Data.npz')
train_Resnet50 = bottleneck_features['train']
valid_Resnet50 = bottleneck_features['valid']
test_Resnet50 = bottleneck_features['test']
(IMPLEMENTATION) Model Architecture¶
Create a CNN to classify dog breed. At the end of your code cell block, summarize the layers of your model by executing the line:
<your model's name>.summary()
Question 5: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. Describe why you think the architecture is suitable for the current problem.
Answer: Because our dog image dataset is very similar to part of the Imagenet dataset, but much smaller, the images making up the new dataset and the dog subset of Imagenet have similar features at a low level (e.g. edges), at an intermediate level (e.g. shapes), and at a high level (e.g. legs, tail, teeth, tongue, etc.). Consequently, it makes sense to exploit most or all of the pretrained neural network layers, as they already contain information which is relevant to the new, smaller, dataset. Therefore, as was done above with the VGG16 network, my model simply replaces the original network's last fully connected layer (the softmax layer) with a new one whose number of nodes matches the number of classes to be predicted for the task at hand (133 instead of 1000). In order to take advantage of the information learnt by the Resnet50 network on Imagenet, all the weights learnt by the pretrained network are frozen, and the new network only learns the weights of the new fully connected layer, which are initialised at random values. The CNN trained from scratch in Step 3 had to learn its parameters on the small dataset alone; in contrast, the pretrained CNN had access to a much larger dataset, which gives far more scope to learn relevant patterns. Also, my CNN from Step 3 is much smaller than the pretrained CNN: ResNet50 has many more layers, and also uses a technique allowing it effectively to tackle the vanishing gradient problem; this setting makes learning more powerful and more effective than my CNN. --- On a separate note, I noticed that compiling the model with Adam as an optimiser gives better results (test accuracy) than with RMSprop.
### TODO: Define your architecture.
Resnet50_model = Sequential()
Resnet50_model.add(GlobalAveragePooling2D(input_shape=train_Resnet50.shape[1:]))
Resnet50_model.add(Dense(133, activation='softmax'))
Resnet50_model.summary()
(IMPLEMENTATION) Compile the Model¶
### TODO: Compile the model.
# Resnet50_model.compile(loss='categorical_crossentropy',
# optimizer='rmsprop',
# metrics=['accuracy'])
Resnet50_model.compile(loss='categorical_crossentropy',
optimizer='adam',
metrics=['accuracy'])
(IMPLEMENTATION) Train the Model¶
Train your model in the code cell below. Use model checkpointing to save the model that attains the best validation loss.
You are welcome to augment the training data, but this is not a requirement.
### TODO: Train the model.
checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.Resnet50.hdf5',
verbose=1, save_best_only=True)
Resnet50_model.fit(train_Resnet50, train_targets,
validation_data=(valid_Resnet50, valid_targets),
epochs=100, batch_size=20,
callbacks=[checkpointer, early_stopping_monitor], verbose=1)
(IMPLEMENTATION) Load the Model with the Best Validation Loss¶
### TODO: Load the model weights with the best validation loss.
Resnet50_model.load_weights('saved_models/weights.best.Resnet50.hdf5')
(IMPLEMENTATION) Test the Model¶
Try out your model on the test dataset of dog images. Ensure that your test accuracy is greater than 60%.
### TODO: Calculate classification accuracy on the test dataset.
# get index of predicted dog breed for each image in test set
Resnet50_predictions = [np.argmax(Resnet50_model.predict(np.expand_dims(feature,
axis=0))) \
for feature in test_Resnet50]
# report test accuracy
test_accuracy = 100 * np.sum(np.array(Resnet50_predictions) ==
np.argmax(test_targets, axis=1)) / len(Resnet50_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
(IMPLEMENTATION) Predict Dog Breed with the Model¶
Write a function that takes an image path as input and returns the dog breed (Affenpinscher, Afghan_hound, etc) that is predicted by your model.
Similar to the analogous function in Step 5, your function should have three steps:
- Extract the bottleneck features corresponding to the chosen CNN model.
- Supply the bottleneck features as input to the model to return the predicted vector. Note that the argmax of this prediction vector gives the index of the predicted dog breed.
- Use the
dog_namesarray defined in Step 0 of this notebook to return the corresponding breed.
The functions to extract the bottleneck features can be found in extract_bottleneck_features.py, and they have been imported in an earlier code cell. To obtain the bottleneck features corresponding to your chosen CNN architecture, you need to use the function
extract_{network}
where {network}, in the above filename, should be one of VGG19, Resnet50, InceptionV3, or Xception.
### TODO: Write a function that takes a path to an image as input
### and returns the dog breed that is predicted by the model.
def Resnet50_predict_breed(img_path):
# extract bottleneck features
bottleneck_feature = extract_Resnet50(path_to_tensor(img_path))
# obtain predicted vector
predicted_vector = Resnet50_model.predict(bottleneck_feature)
# return dog breed that is predicted by the model
return dog_names[np.argmax(predicted_vector)]
Step 6: Write your Algorithm¶
Write an algorithm that accepts a file path to an image and first determines whether the image contains a human, dog, or neither. Then,
- if a dog is detected in the image, return the predicted breed.
- if a human is detected in the image, return the resembling dog breed.
- if neither is detected in the image, provide output that indicates an error.
You are welcome to write your own functions for detecting humans and dogs in images, but feel free to use the face_detector and dog_detector functions developed above. You are required to use your CNN from Step 5 to predict dog breed.
Some sample output for our algorithm is provided below, but feel free to design your own user experience!
(IMPLEMENTATION) Write your Algorithm¶
### TODO: Write your algorithm.
### Feel free to use as many code cells as needed.
# Write an algorithm that accepts a file path to an image and first determines whether the image contains a human, dog, or neither. Then,
# if a dog is detected in the image, return the predicted breed.
# if a human is detected in the image, return the resembling dog breed.
# if neither is detected in the image, provide output that indicates an error.
def print_hello(species):
print("Hello, {}!".format(species))
def print_newlines():
print()
print("====================")
print()
def display_img(path_to_img):
img = cv2.imread(path_to_img)
img_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
plt.imshow(img_rgb)
plt.show()
import os
import re
def get_dog_img_title(img_path):
# "dogImages/test/005.Alaskan_malamute/Alaskan_malamute_00309.jpg" >
# "Alaskan_malamute_00309.jpg"
img_title = os.path.basename(img_path)
# "Alaskan_malamute_00309.jpg" > "Alaskan_malamute"
try:
# "Alaskan_malamute_00309.jpg" > "Alaskan_malamute_"
found = re.search('([_A-Za-z]+).*', img_title).group(1)
# "Alaskan_malamute_" > "Alaskan_malamute"
if found.endswith('_'):
found = found[:-1]
except AttributeError:
found = ''
return found
# Putting it all together: the required algorithm
def predict_dog(img_path):
display_img(img_path)
is_dog = dog_detector(img_path)
is_face = face_detector(img_path)
# If a dog is detected in the image
if is_dog:
print_hello("dog")
print("Your predicted breed is...")
# If no dog is detected, but a human face is
elif is_face:
print_hello("human")
print("You look like a...")
# If neither a dog nor a human face is detected in the image
else:
print("Are you sure your image shows a dog or a human?")
print("How about trying another photo?")
print_newlines()
return
# If dog or human face
breed = Resnet50_predict_breed(img_path)
print(breed)
# Display a sample image of the predicted dog breed.
print("See for yourself!")
sample_breed_imgs = [file_name for file_name in train_files if breed in file_name]
sample_breed_img = random.choice(sample_breed_imgs)
display_img(sample_breed_img)
# Just for testing, if it's a dog check the title of the image
# to see if the actual breed matches our prediction.
# That's because I am not familiar with most dog breeds, and I'd like to assess
# how well we are doing. And I know that in my sample dataset file titles contain
# either the breed (if the image is of a dog) or the main object shown in the image.
# This should probably not be part of the actual app.
if is_dog:
print("Am I right? Your image is called {}".format(get_dog_img_title(img_path)))
print_newlines()
Step 7: Test Your Algorithm¶
In this section, you will take your new algorithm for a spin! What kind of dog does the algorithm think that you look like? If you have a dog, does it predict your dog's breed accurately? If you have a cat, does it mistakenly think that your cat is a dog?
(IMPLEMENTATION) Test Your Algorithm on Sample Images!¶
Test your algorithm at least six images on your computer. Feel free to use any images you like. Use at least two human and two dog images.
Question 6: Is the output better than you expected :) ? Or worse :( ? Provide at least three possible points of improvement for your algorithm.
Answer: Because my dog breed predictor will make a prediction no matter what the input - a landscape, a human face, or any other object -, it is funny to try to relate the predictor's outcome with features in the input image (e.g. why is a building predicted to be a Papillon dog, or a beach a Dalmatian?). I experimented a bit with this before filtering out non-dog and non-human face images, and it was quite amusing to find an actual resemblance between a dog and an inanimate object. My algorithm takes advantage of this behaviour in some cases (when the input is a human face), but avoids having to handle other types of input (predicting a dog breed based on a human face may seem funny, but doing so based on other types of object is a window into the weakness of our method, which is, by itself, unable to guarantee its prediction is really applicable to its input). This is why the breed classification is preceded by a binary dog/non-dog and face/no-face classification step. So the computation proceeds in two steps, not one as it may seem to the user:
- Filter out irrelevant input images, i.e. the ones I have beforehand decided I do not want to handle (non-dog and non-human face).
- Predict dog breed for the remaining images (which would have been done for those filtered out too, failing step 1).
This is why it is essential that the dog/face detector works well, or we will get some undesired results (dog breed predictions for things other than dogs or human faces).
On my small 26-image sample, the breed prediction is rather good for dog images (the classifier did get over 80 percent accuracy on the test set), and not unreasonable for human faces. The dog detector seems to work pretty well too: it does not seem to mistake other animals for dogs (there are some chamois in my images, and they are not mistaken for dogs). The face detector's performance is a bit lower: it does mistake a cat for a human and detect a human whose face is hidden. So overall I would say I expected worse performance, though I think more testing is required on borderline or tricky cases.
Here are a few mistakes I have noted on the sample set:
- Surprisingly enough, a cat is mistaken for a human, not for a dog! This confusion may be due to the fact that the human detection is based solely on the face, and the features used for human face detection are also found in this cat's face.
- There is also a minor mistake in the breed predicted for a dog whose face is hidden. But the real breed and the predicted one are really very similar: bodies are virtually indistinguishable, and heads are a few subtle differences apart. So I would not count this as a mistake really. I would even say the result is surprisingly good given that the dog's head is not visible.
- On one photo (the one with the beach) the presence of a dog goes undetected. True, the silhouette is very small. Is the dog detector's scale invariance too limited to handle such cases?
- Another interesting behaviour is the way the program handles a photo of a human seen from behind touching a stone statue of an imaginary animal. In the first step, the image is labelled as showing a human, not a dog. However, in step 2 the dog breed classification is applied to the statue, not to the human (whose face is not even visible). This is a clear indication that the classification is performed in two totally disjoined steps.
And here are three possible points of improvement for my algorithm:
- Make the dog and human face detectors not exclusive. Right now, if a dog is detected the human detector is not triggered too, even if there is also a human in the image. This would enable the program to handle multiple target objects. If both a dog and a human face, or more than one dog or face, are detected, then the program should predict a dog breed for each.
- For the same purpose, add object localisation to the dog and face detectors. This way, when more than one target object is detected, the dog breed predictor should run on each one, and give a prediction for each relevant location.
- Improve the human face detector. Right now it gives false positives.
## TODO: Execute your algorithm from Step 6 on
## at least 6 images on your computer.
## Feel free to use as many code cells as needed.
# test
# predict_dog("dogImages/test/005.Alaskan_malamute/Alaskan_malamute_00309.jpg")
# test
# predict_dog("lfw/Aaron_Eckhart/Aaron_Eckhart_0001.jpg")
Create a set of sample images.
def pick_imgs(list_of_paths, how_many_to_pick, seed):
np.random.seed(seed)
return np.random.choice(list_of_paths, how_many_to_pick)
import os
# Dog images from the dog test set.
dogs = pick_imgs(test_files, 3, 555)
# Human faces from the human dataset.
people = pick_imgs(human_files, 5, 888)
# A mixture of dogs, faces, but mostly landscapes, from my own images.
my_imgs = [os.path.join('myImages', img) for img in os.listdir('myImages/')]
sample_imgs = np.concatenate((dogs, people, my_imgs))
np.random.seed(789)
np.random.shuffle(sample_imgs)
sample_imgs
# How many sample images?
sample_imgs.shape
Test the algorithm on the sample images.
for img_path in sample_imgs:
predict_dog(img_path)
